Holter triage ambulatory ECG analysis. Accuracy and time efficiency

Triage ambulatory electrocardiographic (ECG) analysis permits relatively unskilled office workers to submit 24-hour ambulatory ECG Holter tapes to an automatic instrument (model 563, Del Mar Avionics, Irvine, CA) for interpretation. The instrument system "triages" what it is capable of automatically interpreting and rejects those tapes (with high ventricular arrhythmia density) requiring thorough analysis. Nevertheless, a trained cardiovascular technician ultimately edits what is accepted for analysis. This study examined the clinical validity of one manufacturer's triage instrumentation with regard to accuracy and time efficiency for interpreting ventricular arrhythmia. A database of 50 Holter tapes stratified for frequency of ventricular ectopic beats (VEBs) was examined by triage, conventional, and full-disclosure hand-count Holter analysis. Half of the tapes were found to be automatically analyzable by the triage method. Comparison of the VEB accuracy of triage versus conventional analysis using the full-disclosure hand count as the standard showed that triage analysis overall appeared as accurate as conventional Holter analysis but had limitations in detecting ventricular tachycardia (VT) runs. Overall sensitivity, positive predictive accuracy, and false positive rate for the triage ambulatory ECG analysis were 96, 99, and 0.9%, respectively, for isolated VEBs, 92, 93, and 7%, respectively, for ventricular couplets, and 48, 93, and 7%, respectively, for VT. Error in VT detection by triage analysis occurred on a single tape. Of the remaining 11 tapes containing VT runs, accuracy was significantly increased, with a sensitivity of 86%, positive predictive accuracy of 90%, and false positive rate of 10%. Stopwatch-recorded time efficiency was carefully logged during both triage and conventional ambulatory ECG analysis and divided into five time phases: secretarial, machine, analysis, editing, and total time. Triage analysis was significantly (P < .05) more time efficient for the machine, analysis, and total time phases and required significantly more time in the editing phases. The triage analysis was limited by the inability of the automatic methodology to detect a specific VT waveform contained on one tape. This result which caused substantial loss of sensitivity for VT, emphasizes the necessity for careful quality control and editing of all automatic Holter analysis methods.